Medical terminologies for patients - SemanticHealthNet 288408... · Medical terminologies for...

Medical terminologies for patients

Elena Cardillo

Institute of Informatics and Telematics UOS Rende (CS), National Research Council, Consenza, Itally

Annex 1 to SHN Work Package 3 Deliverable D3.3

Final version, March 29, 2015

Medical terminologies for patients. Elena Cardillo

1

Document description

Deliverable: Annex 1 to SHN WP3 Deliverable D3.3

Publishable summary:

In this Annex, the state of the art of consumer-oriented medical terminologies is described, together with the tools to evaluate readability of medical texts. An approach is presented to the development of a multilingual consumer-oriented vocabulary, covering the most prevalent concepts in medical communication. Finally, integration of such a vocabulary into health information systems and linking to multiple pertinent international reference terminologies (International Classification of Primary Care, International Classification of Diseases, SNOMED-CT) is discussed.

Status: Final Version

Version: 1.5

Public: x Yes

Deadline: May 29, 2015

Contact: Elena Cardillo [email protected] Robert Vander Stichele [email protected]

Editors: Elena Cardillo

Table of content 1 Introduction ............................................................................................................................................... 2

2 The communication gap in medicine ........................................................................................................ 4

3 State of the art in the development of consumer-oriented medical terminologies ................................. 5

3.1 Consumer-oriented Vocabularies in the medical domain ................................................................. 5

3.2 Techniques and tools for evaluate and enhance readability of medical information ...................... 7

4 Combined approach for the development of consumer-oriented medical vocabularies ......................... 8

4.1 Generation of the vocabulary ............................................................................................................ 8

4.2 Integration of the vocabulary with standardized terminologies and classification systems .......... 10

4.3 Evaluation of consumer-oriented vocabularies and application in healthcare information systems

12

5 Conclusions .............................................................................................................................................. 14

References ..................................................................................................................................................... 15

Note :

This annex was commissioned by Prof. Dr. R. Vander Stichele, Workpackage Leader of SemanticHealth Net

WP3, to Elena Cardillo, PhD, Institute of Informatics and Telematics UOS Rende (CS), National Research

Council, lo, PhD, Institute of Informatics and Telematics UOS Rende (CS), National Research Council, Via P.

Bucci, 17B, 7 floor, 87036, Rende, Cosenza, Italy

Email: [email protected] Tel. : +390984494957

mailto:[email protected]




2

1 Introduction

During the last twenty years efforts in research and innovation in the field of Information

Technologies for the biomedical domain and in particular for electronic Healthcare (eHealth) have

defined new methodologies and standards to improve the management of healthcare data and

processes of care. The integration of all the different information systems which need to exchange

these data, and finally the creation of healthcare systems useful both for healthcare professionals

(Electronic Health Records - EHRs) and for healthcare consumers (Personal Health Records -

PHRs1) have also been improved. Thanks to the institutional measures and to the European

eGovernment Action Plan 2011-2015, some European countries have started integrated

management of patients' healthcare data, with the creation and the distribution of a federated

solution of “Electronic Health Record” to be aligned with the international scenario.

To reach this aim, many projects and localized initiatives have been launched and infrastructures

useful to support this solution have been created. These focus in particular on the Patient

Summary and on electronic Prescription, searching for the alignment of medical encodings related

to these two processes. Few efforts and contributions, on the other hand, are related to other

important building blocks such as the Personal Health Records (PHRs), for the autonomous

management and organization of healthcare data by the patient and on integrated medical

vocabulary for the patient.

Standardized methodologies are developed to encode, retrieve, represent and integrate

healthcare data, terminologies and classification systems by and for healthcare consumers, to help

them define their personal and familiar clinical history and to ease access to clinical data and

communication with professionals. The creation of personal health records meets the international

challenge of “Patient Empowerment”, giving more power to consumers for managing and

organizing their own healthcare data. Patients want to be able to review and contribute to their

records, to include their perspectives and priority symptoms for discussion and shared decision

making with clinicians. To this end they need support to create the best possible clinical

documentation and help with data entry in order to avoid degradation of information. This means

investing in interface terminologies, dealing with multiple languages, and with the difference

between lay language and medical jargon. Standard international classifications or terminologies

may be needed to bring a real semantic interoperability, but their use may too complex even for

domain expert, and certainly for patients. It is necessary to help healthcare consumers and all

possible end-users (patients, physicians, nurses, etc.) to better understand the available

terminologies. In order to allow high quality data entry and useful applications, medical

terminologies must reflect the words and phrases of both intended users: clinicians on one hand

and patients on the other.

To mitigate the linguistic gap between the lay language adopted by healthcare consumers2 and the

specialized, technical language of physicians and other healthcare providers, lay terminologies or

vocabularies need to be created. Using these systems would allow easier and more efficient

management and interpretation of patients’ healthcare data, and better understanding of medical

reports by consumers. In addition, physicians could import terms expressed by the patients into

1 PHRs (Personal Health Records) are electronic health record managed by the patient, generally available on the web, that differ from EHRs since

they are updated, and integrated directly by the patient with data on their clinical history and it is usually used to access to their clinical reports or test results, or for the self-monitoring of specific diseases conditions.

2 The terms patients and healthcare consumers used to identify any actual or potential recipient of health care.


3

their EHRs, automatically encoded (e.g. symptoms, administrative procedure requests, adverse

events, etc.).

A first step in this direction is the creation of PHR models which could be fully integrated in the

EHR framework. A further step is needed for the identification of a contextualized lay language

used by consumers for expressing healthcare concepts and consequently for the creation of

consumer-oriented lexicons or vocabularies which implies the extraction of semantic and linguistic

correspondences between lay language and international standardized medical classification

systems, nomenclatures, and thesauri (ICD-10, ICD9-CM, ICPC-2, LOINC, SNOMED CT, UMLS,

and MeSH) used all over Europe and in many other foreign countries for the medical

documentation and coding and for medical information retrieval. The use of these lay terminologies

and their interoperability to standardized terminologies can assure consumers easy access to and

interpretation of their data, also in case of emergency, for example during travels abroad.

Concerning this last point, but also referring to those patients such as immigrants which have

difficulties first of all with the hosting language and secondly with the technical jargon used in the

health care setting, the multilingual aspect needs to be considered by providing the development of

consumer-oriented medical vocabularies in various languages. This would be an added value

together with semantic interoperability that will facilitate cross-border care.

Finally, the use of new technologies and languages, such as the one used in the Semantic Web [4]

to represent these terminological resources and to model the knowledge included in healthcare

information systems (e.g. EHRs and PHRs) will facilitate a series of knowledge services and

automatic reasoning on the patient clinical data which will allow an easier and more efficient

management (by clinicians) or self-management (by patients) and interpretation of healthcare data

as well as an easier and quicker retrieval of medical information.


4

2 The communication gap in medicine

Electronic Health Records necessitate the integration of medical terminologies and coding systems

that ease the registration of clinical data by the GPs or other healthcare professionals and that

allow for the structured and interoperable reporting of clinical data. Such resources are

characterized by a physician-oriented technical language. In the past years, steps forward were

taken in the field of e-Health and so-called “Patient Empowerment”. Patients or more generally

healthcare consumers have taken an active role, both in the consultation of medical information

online (thanks to the proliferation of dedicated websites) and in the access to and management of

their healthcare data through the use of PHRs, available on the web, on their mobile or tablet. It is

evident that consumers need support to read, interpret, and manage their medical data.

A solution, at least from the patient perspective, would be a simplification of the medical language

and consequently a cleaning process of the language itself, through the creation of a terminology

composed of well-defined and rigorously applied words. This does not imply, however, the abolition

of specialized terms, because no specific language can do without its lexical background.

Moreover, the medical language is overloaded with obsolete and archaic terms, eponyms, multiple

synonyms, and semantic ambiguities [31].

Too often the professional inclination to use highly technical terms makes the patient

uncomfortable, implying a strong effort in order to understand. Physicians and healthcare operators

do not talk “with” the patient but “to” the patient [2]. Patients need clear and understandable

language to communicate effectively [34]. These statements become more relevant when

considering the communication (not mediated by the physicians) between patients and the recent

tele-health applications.

A solution to the problem is the creation of medical vocabularies, terminologies or ontologies,

specifically designed for patients [6]. The challenge is to map these terminological resources for

patients to the specialized medical terms.


5

3 State of the art in the development of consumer-oriented medical

terminologies

3.1 Consumer-oriented Vocabularies in the medical domain

Over the past 20 years researchers have worked on the development of lexical resources and

terminologies that reflect the way healthcare consumers express and think about health related

concepts.

An early example of a consumer-oriented medical terminology which considers multilingualism was

commissioned by the European Commission in 1994. The Multilingual Glossary of Popular and

Technical Medical Terms contained both lay and technical terms and expressions in 9 European

languages and was limited to the terminology used in medication leaflets. It is, in fact, composed

of the 1,400 most frequent technical terms used in drugs package inserts, with corresponding lay

terms or definitions in English, Danish, German, Spanish, Italian, Dutch, Portuguese and Greek

[21].

Later Soergel and Tze defined a methodology for the development of a common medical

vocabulary, namely Consumer Medical Vocabulary, mapped to two different resources: (i) an

intermediate vocabulary named Mediator Medical Vocabulary used by health care operators such

as nurses who mediate between patients and clinicians, and (ii) a specialized medical vocabulary

Mediator Medical Vocabulary used by professionals [43]. In this initiative linked lay and technical

terms have also been mapped to the UMLS Metathesaurus to find synonyms and quasi-synonyms,

and an intermediate layer has been created to interpret and mediate among the different types of

vocabularies. A large number of common expressions (hundreds of thousands of tokens) were

examined, leading to the discovery that between 20% and 50% of the lay expressions was not

represented in the specialized medical vocabularies [25].

One of the major international efforts in this regard is the Consumer Health Vocabulary Initiative

(CHV Initiative), launched in 2006 at the Brigham and Women’s Hospital, Harvard Medical School

for the development of the Open Access Collaborative Consumer Health Vocabulary (OAC CHV).

It is a consumer-oriented medical vocabulary for English defined as: “a collection of forms used in

health-oriented communication for a particular task or need […] by a substantial percentage of

consumers from a specific discourse group and the relationship of the forms to professional

concepts” [47]. In particular, the CHV includes common medical terms and their synonyms in

multiple medical subdomains. In 2009 the CHV was officially incorporated in the Unified Medical

Language System (UMLS) Metathesaurus. The lay terms of the CHV were linked to technical

medical concepts (e.g. Shortness of breath linked to Dyspnea).

This type of vocabulary can have three possible bridging roles between consumers and health

applications or information systems:

Information Retrieval, because CHV facilitates automated mapping of consumer-entered

queries to technical terms, producing better search results;

Medical Records, since medical records and test results are nowadays available to

patients, they frequently contain jargon, so a CHV can represent these terms with

consumer-understandable names to help patients better interpret the medical concept;


6

Health Care Applications, where patients may enter consumer expressions such as “nose

bleeding” or “cluster headache”, receiving help via an integrated CHV, which would

facilitate automated mapping of these expressions to technical concepts (in this case

“epistaxis” and “histamine cephalalgia”) enabling consequent analysis and response.

In some cases, these vocabularies were applied in concrete use cases [27, 46].

In the United States context, many initiatives promoted by for-profit companies can be found. An

example is the Consumer Health Terminology Thesaurus, developed by WellMed Inc., which is

based on SNOMED (Systematized Nomenclature of Medicine) and contains more than 20,000

terms familiar to the patient, including many cases of dialectical and cultural lexical variants [51]

[35].

In Europe, the Italian Consumer Medical Vocabulary (ICMV) was created. In line with the OAC

CHV, it provides two main contributions: the creation of an Italian medical vocabulary oriented to

consumers and patients developed by applying an hybrid methodology of knowledge acquisition

and terminology extraction validated by domain experts [9]; and the integration of this resource

with some specialized medical terms in the UMLS Metathesaurus (ICPC-2, ICD-10, SNOMED CT,

LOINC) used by healthcare professionals in primary and secondary care, by using Semantic Web

technologies and languages3 [8].

One of the most recent attempts in building lay terminologies is the Mayo Consumer Health

Vocabulary (MCV), a taxonomy of approximately 5,000 consumer health terms and concepts

partially mapped to SNOMED CT and ICD-9 [40]. The authors here also developed text-mining

techniques to expand its coverage by integrating disease concepts from UMLS as well as non-

genetic (from deCODEme4) and genetic (from GeneWiki+5 and PharmGKB6) risk factors to

diseases. A comprehensive review of the literature from different databases (e.g. PubMed

MEDLINE, CINAHL) and information sources (Library and Information Science Abstracts, and

Library Literature) about consumer and patient language, and controlled vocabularies showed that

consumer contributions to controlled vocabulary appear to be seriously under-researched inside

and outside of health care [41].

Other works have been focused on the extension of consumer health vocabularies for specific

medicine subdomains [36], where terms and the expressions used by lay persons speaking French

to talk about breast cancer are identified and organized in a concept-based terminology.

A computer assisted update (CAU) system of the open access and collaborative (OAC) CHV is

presented in [15], as a system consisted of three main parts (i.e. a Web crawler, an HTML parser,

and a candidate term filter) that identifies new candidate terms from live corpora for inclusion in the

(OAC) CHV. The CAU system was applied, for evaluation, to the health-related social network

website PatientsLikeMe.com identifying 237 valid terms not yet included in the OAC CHV or in

UMLS, among 774 candidate terms selected by the term filter from 300 crawled webpages.

3 In particular the use of languages such as RDF and OWL for the formal representation of the terminological resources and of the vocabulary

itself, and the use of SPARQL as query language for the extraction of the semantic correspondences. Finally, the use of collaborative tools such as

Semantic Media Wiki for the terminology acquisition directly from users and for the clinical mapping by domain experts, as well as for the final

publication of the Vocabulary. 4 http://www.gene-tests.org/decodeme 5 http://genewikiplus.org/wiki/Main_Page 6 https://www.pharmgkb.org/


7

3.2 Techniques and tools for evaluate and enhance readability of medical

information

The reviewed literature suggests that while giving patients access to medical documents has many

benefits, lay people have difficulty in understanding medical information and in most cases this

causes problems that only with the help of supporting tools can be solved. This requires first of all

in-depth understanding of the nature and causes of comprehension errors that lay people make

when dealing with clinical documents (e.g. misunderstanding of clinical concepts, misreporting of

physician’s findings, confusion or misspelling of clinical terms) [26]. In recent years research has

focused to this end on the assessment of the readability of clinical documents as well as

techniques useful to improve readability and guarantee the easier access by lay people to medical

information. In [53] the problem has been approached by contrasting the ‘readability’ of two types

of clinical documents: referral letters (76,012) vs. other genres of narrative clinician notes

(2,118,463), using as a baseline a corpus of MedlinePlus articles—exemplars of fine patient

education materials crafted for lay audiences. The readability has been quantified using three

different measures: Flesch-Kincaid Grade Level (FKGL); Simple Measure of Gobbledygook

(SMOG); Gunning-Fog Index (GFI). This work presented some limitations in the medical

documents analyzed by the authors, retrieved from a single institution and patient care service

(hematology/oncology), with the implication of the difficulty in the generalization of the results to

medical content produced by other institutions or care setting. Another limitation recognized by the

authors is the use of only computational measures to estimate readability. Other works are present

in the literature aimed at the assessment of readability of clinical texts for lay people (see for

instance [42], and [49]), but only few of them propose text simplification methods and tools to

improve patients' electronic health record comprehension. One example is described in [24], where

a simplification tool has been developed to simplify health information, that addresses semantic

difficulty by substituting difficult terms with lay synonyms or through the use of hierarchically and/or

semantically related terms (e.g. hyponyms or hyperonyms). The described tool also simplifies long

clinical sentences by splitting them into shorter grammatical sentences, and has been tested to

simplify electronic medical records and journal articles with good outcomes (e.g. for electronic

medical records a statistically significant improvement has been shown in the cloze test score from

35.8% to 43.6%). Another good example of a text simplification tool is the BioNLP system NoteAid

developed by [38], a system composed of a Concept Identifier module that performs typical NLP

tasks and also matches concepts in other resources and a Definition Locator module that looks for

concepts definitions from UMLS, MedlinePlus and Wikipedia. This system integrated in patients’

electronic health records automatically recognizes medical concepts and links these concepts with

consumer oriented, simplified definitions from external resources. An evaluation of the system

showed that Wikipedia significantly improves EHR note readability, while MedlinePlus and the

UMLS need to improve their content coverage for consumer health information in order to be

useful as resources for NoteAid. A similar method has been proposed in [33], by computing in this

case term familiarity to help estimate text difficulty.

All these studies highlight the necessity of informatics support tools to be used by health care

professionals, to make clinical information understandable to patients or on the contrary used by

healthcare consumers to better understand their clinical documents or healthcare information on

the web.


8

4 Combined approach for the development of consumer-oriented

medical vocabularies

Most of the approaches reviewed in this report share some peculiar steps that need to be

performed when building vocabularies or terminologies oriented towards healthcare consumers. In

particular it is possible to identify 4 main steps for the development of consumer medical

vocabularies (CMV):

1) Identification of consumer-friendly terms used to indicate medical concepts in daily life and

during their encounters with health care professionals (in particular to express symptoms

and complaints, medical procedures and diseases), that can be performed by means of

different elicitation techniques and usability studies and using automatic term extraction

techniques from different corpora or websites oriented towards consumers;

2) Review of the consumer-friendly terms and selection of the candidate terms for the CMV;

3) Mapping of the selected terms to the corresponding terms/concepts in standardized

medical terminologies, classification or ontologies used in the domain of application (in

particular primary and secondary care) by means of a semi-automatic approach, including

validation of the mappings by domain experts;

4) Evaluation of the feasibility of the vocabulary by means of its application within personal

health records or its application to search engines. A brief description of these tasks is

given in the next sections, in particular steps 1) and 2) in Section 4.1.; step 3) in Section

4.2. and step 4) in Section 4.3.

4.1 Generation of the vocabulary

One of the crucial steps in developing consumer medical vocabularies (CMV) is the identification of

consumer health expressions. In order to identify consumer-friendly terms, usability studies can be

performed, as shown in [22], where a usability study of the patient-friendly terms used in an

ambulatory electronic medical record and associated patient web portal has been carried out

investigating the usage patterns of consumer health vocabulary and evaluating the mapping to

controlled terminologies used in the electronic medical records (e.g. UMLS). Other usability studies

have been executed by [29] to find out the differences between consumer and medical vocabulary.

When carrying out these usability studies it has to be considered that the use of patient-friendly

terms for some types of medical concepts would not always help to bridge the language gap

between providers and consumers. In fact, considering diagnoses, the professional terms are used

more frequently than their patient-friendly counterparts, typically in cases where the professional

terms are more simple or common than the patient-friendly terms (e.g. in the case of Diabetes,

Hypothyroidism, etc.). Consequently a low level of usability of lay terms for diagnoses has not to be

seen as a negative result. What is helpful in this case is the identification of lay expressions or

descriptions to clarify the meaning of a diagnosis.

Generally, combined approaches of semi-automatic analysis of text corpora and manual revisions

performed by domain experts are preferable for the identification of lay linguistic forms used by

healthcare consumers [48].

It needs to be highlighted that most of the methods used by existing studies in identifying

consumer health expressions involve human efforts (e.g. by organizing interviews or focus groups

with different samples of people), which is very time-consuming. Frequent is also the use of clinical

guidelines as terminological resources, but being professional-oriented do not seem completely


9

adequate to the objective of this task. These issues can be reduced by considering as a

terminological source online health social media sites that provide consumers with healthcare

information sources as well as communication platforms for social interactions such as discussion

forums and online social groups. These platforms collect an enormous amount of evolving

consumer-contributed healthcare content. A great deal of information can be found on these social

media sites before they are reflected to health professionals or recorded by some other means

(health consumer opinions on new medical treatments, drugs, vaccine, etc.). So it turns out to be a

great resource to harvest the timely consumer health expressions, which are not available through

other channels. As these sources include both patients’ comments and questions and answers

posted by healthcare professionals, it is important to find methods to categorize posts produced by

patients and those produced by health professionals in order to extract exactly consumer-friendly

terms. A similar study can be found in [19] and [Error! Reference source not found.], where a

supervised approach based on n-grams (vocabulary), emotion markers, uncertainty markers and

misspellings has been evaluated to distinguish the two categories of posts on a French health

forum (AlloDocteurs.fr).

One example of the automatic identification of relevant lay terms or expressions from consumer-

contributed content on the web is the execution of co-occurrence analyses on consumer-based

corpora such as messages on forums dedicated to specific medical subdomains, as done in [30],

where relevant lay terms related to Adverse Drug Reactions (ADRs) have been extracted from a

corpus of 120,393 discussion messages on a forum. Another example is the hybrid approach

proposed in [9], that combines traditional knowledge acquisition techniques with the automatic

collection of consumer-friendly terms from various sources on the web, in particular forum postings

written by healthcare consumers on medical-consultation websites (i.e. medicitalia.it7), by applying

NLP techniques and tools for term extraction, parsing, tagging and normalization. By means of

NLP tools for term extraction it is possible to detect single and multiword medical terms and a basic

semantic structure defining relations between the extracted terms (BT, NT, RT). Some of these

tools (e.g. Text 2 Knowledge –T2K, or the tool used in the Terminology Extraction for Semantic

Interoperability and Standardization project - TExSIS) developed specific adaptation modules to

the biomedical domain [14, 32], being in this way more reliable in terms of precision and recall and

for the extraction of semantic relations between terms. In fact, extracted single and multiword

medical terms can be structured into fragments of taxonomical chains reconstructed from the

internal linguistic structure of the terms itself (e.g. corneal abrasion is automatically associated with

the relation IS-A to the term abrasion. Clusters of semantically related terms (RT) can be inferred

through dynamic distributionally-based similarity measures using a context-sensitive notion of

semantic similarity (computed with respect to the most relevant co-occurring heads). Doing so it is

possible to extract relations between the term “contusion” and the terms sprain, and injury.

Concerning the selection of the candidate terms to be included in a CMV, it can be divided into two

main tasks: (i) a statistical analysis based on term frequency and on the degree of familiarity of the

extracted terms (which represents the level of understandability and use of a certain medical term

for healthcare consumers) possibly assigned by a sample of healthcare consumers; and (ii) a

clinical and semantic review of the extracted terms performed both by one or more domain experts

(e.g. physicians, nurses and pharmacists) and by terminologists. The manual review by physicians

serves principally for quality assurance, while the review by terminologists can be useful to find

mistakes and incongruities in categorization and synonymy. As mentioned in [26], possible and

categorizing types of mismatches from automated mechanisms include: misspellings (e.g., “erpes”

7 http://www.medicitalia.it/


10

instead of “herpes” or “celebral ictus” instead of “cerebral ictus”), truncations (e.g., “Down” for

“Down Syndrome”), acronyms, abbreviations and word fragments.

The review step include a conceptualization of the extracted terms and consequently the

association to each term of a standardized definition in order to make explicit their meaning as well

as perform a general categorization of the concepts (e.g. is the concept a disease, a symptom, an

anatomical concept, etc.). This task is important to disambiguate such terms that can have different

meaning, so homonyms (terms that have the same orthographical form and phonology but can

express more than one meaning). In fact, there can be terms such as the term “mark” or “spot”

used in medicine to indicate a visible sign on the skin (defined respectively as “a visible impression

on a surface, as a line or spot”, and “a small blemish or other mark on the skin”), but also having

the other “non-medical” meanings (e.g. respectively “a symbol used in writing or printing”, a

punctuation mark, “a position in an organization or hierarchy”, etc.). It is important, then to

recognize the medical meaning of the concepts in this phase in order to allow ahead a semantically

correct alignment of the selected patient-friendly terms and those included in international medical

classifications or terminologies.

Some of the studies mentioned in Section 3.1., in order to find standardized definition of the

patient-friendly terms, performed automated extraction of the definitions, in English, from the UMLS

Metathesaurus (see for instance [47], [43] and [10]), so considering the so-called “Concept Unique

Identifier” (CUI), a code that represents a concept in the UMLS Metathesaurus. The possible

‘senses’ of a term can be considered as the set of CUI’s which list this term as a possible

realisation (e.g. the term trauma in UMLS is a possible realisation of the two concepts: C0043251

Injuries and Wounds in the sense of traumatic injury and C0021501 Physical Trauma. Since in

UMLS this term can be used to express more than one UMLS concept, a disambiguation process

is needed to find out which of its possible senses is actually being used in each particular context

where there term trauma is used [52].

4.2 Integration of the vocabulary with standardized terminologies and

classification systems

After creating the consumer-oriented vocabulary, a crucial step is its semantic integration with

standardized technical and physician-oriented medical terminologies or classification systems. This

consists in finding for each selected lay term the corresponding technical ones in resources that

are widely distributed and used worldwide in the domain of application (e.g. ICD-9-CM, ICD-10,

ICPC-2, SNOMED CT, UMLS, etc.). By means of the mapping process it is possible to reconstruct

the meaning inherent in the lay usage of a term, and consequently to show that compatibility

between lay and professional terms exists on the basis of this deeper meaning, rather than on the

basis of the lexical form. In order to have an integration that could be feasible at a local level, it is

important to map CMV not only with standardized international terminologies and classification

systems but also with reference terminologies and end-user terminologies that represent the

concepts and terms used in the daily practice by physicians in a specific local context. These

terminologies are generally created by extracting concepts and terms directly from EHRs, by

guidelines and also by online medical consultations. Many studies have been published on

analysis of physicians language extracted from EHRs and compared to standardized coding

systems [7] and also on the development and semantic integration of reference and end-user

terminologies (see for instance [28Error! Reference source not found.] and [10]). The

correspondence between consumer-friendly terms in the CMV and the terms in the selected


11

standardized resources can be of four different types: exact mapping between the pairs, when the

lay term has exact correspondence to a term in the other resource and both have the same

meaning; synonymy relation, when the lay term does not exist in the professional resource, but

corresponds to a technical term that denotes the same concept, as in the case of nosebleed and

epistaxis; hyponymy relation, if the lay term is considered as term of inclusion of one or more

concepts in the other resource (e.g. the relation between absence of voice and the broader

concept voice symptom); hypernymy relation, when the lay term includes in its meaning one or

more terms in the other resource (e.g. the relation between bronchitis and chronic bronchitis.

Mappings can be performed manually or automatically. When CMV does not include a high

number of terms and it needs to be mapped only to one standardized resource that is not very

granular either (e.g. ICPC), then manual mapping can be performed. In this case, domain experts

are generally called to manually find a “one-to-one” or “one-to-n” mapping of lay terms with the

corresponding medical concepts in the other resource, to define explicit relationships among them.

On the contrary, in the presence of a large vocabulary and considering to align more than one

standardized resource to the CMV, than an automatic or semi-automatic mapping process is

preferred. Some semi-automatic methods for detecting matching between biomedical vocabularies

are based on similarity functions and measures applied both to single words and n-grams, as

proposed in [37].

In [6] both the approaches have been used. First, a clinical manual mapping between the Italian

Consumer-oriented Medical Vocabulary (ICMV) and the International Classification of Primary

Care, 2nd revision (ICPC-2) has been performed by a sample of general practitioners, who

identified correspondences between Italian lay expressions or terms in the ICMV (e.g. “sentirsi il

cuore in gola” - feel your heart your throat) and the ICPC-2 rubrics (e.g. in this case K04

Palpitazioni - Palpitations). Second, the automatic mapping between the ICMV and selected

international classifications and terminologies in the UMLS Metathesaurus (more precisely

SNOMED CT, ICD-10, MeSH and LOINC), has been executed by using semantic web

technologies [8]. Here, in particular, after the encoding of the ICMV and of the selected

international resources in RDF (Resource Description Framework), the resulted graphs have been

collected in an RDF triple store (e.g. Virtuoso, or Sesame), and SPARQL queries performed on the

stored graphs to extracted semantic mappings between them using ICPC-2 as a pivot to access

UMLS vocabularies. At the end of this process, manual mapping between ICMV and ICPC2 and

the automatic mapping to UMLS vocabularies via ICPC2 was compared, in order to evaluate the

best approach.

The use of Semantic Web technologies has led to promising results in the area of information

integration across heterogeneous resources. For example the work of Bodenreider (2008) can be

mentioned [5], where the Resource Description Framework (RDF) is used for comparing formal

definitions between LOINC and SNOMED CT and in SNOMED CT and the NCI Thesaurus.

However, the major use of these technologies for healthcare has been tested for the formalization

of existent medical terminologies or classification systems in ontologies (one of the latest works in

this sense is the collaborative ontological development of ICD-11 revision, coordinated by the

Stanford Centre for Biomedical Research (BMIR) [45]. These technologies are very useful in

healthcare, especially considering that the treatment of a patient may involve several practitioners

from different healthcare institutes, and that there is an increasing need to access patients'

healthcare records electronically wherever they are stored. Knowledge representation techniques

provide, for example, suggestions on how to manage a patient's condition; tests that have to be


12

carried out; medications or treatment to take into account, etc. In this case ontologies8 become

relevant if integrated into EHRs or PHRs, since they manage an increasing volume of narrative

data, in order to allow: structuring and semantics of the recorded information; and references to

concepts from terminologies such as ICD 10/9 or SNOMED CT [11].

After performing the definition of mappings between consumer-oriented medical vocabularies and

other international medical resources an evaluation of the quality assurance of the mappings

needs to be performed.

The integration of consumer-oriented medical vocabularies with international terminologies and

classification systems is needed to supply knowledge services to support the development of

semantic-based healthcare information systems which implies interchanges with patients and

healthcare consumers in general and consequently to guarantee semantic interoperability.

4.3 Evaluation of consumer-oriented vocabularies and application in

healthcare information systems

Older efforts to improve consumer-friendliness of PHR and EHR information have focused on user

interface design and/or the links to references or educational materials (i.e., infobuttons) [13], [3].

New methods for evaluating the feasibility of consumer-oriented vocabularies consider their

integration in PHRs for supporting healthcare consumers in data entry of medical terms (e.g.

symptoms, diseases, interventions, allergies, and other relevant information of their clinical

history), but above all for supporting them during medical information searching for the

comprehension of their clinical reports, test results, discharge letters, etc., available on their PHR.

In this case the CMV and its integration framework are used as knowledge sources to be queried

when consumers edit terms on their searching panel on their PHR or need to add terms on specific

fields of the application.

PHRs can communicate with physicians’ EHRs, for example to receive medical reports, test

results, and other documents which use specialized medical terminologies (e.g. ICD-9, ICD10) so

having integrated a CMV could give the possibility to foster the readability of data deriving from

EHRs but also to send to physicians PHR content filled out in lay terms and then to translate this in

technical language.

During the last five years, many experimental uses of PHR systems were promoted to improve

patient empowerment. The scientific community launched standardization initiatives to develop

common formats for allow consumers and patient collecting and managing personal healthcare

data and to solve the problem of data interoperability with EHRs and other healthcare information

systems [18]. In some EU countries normative measures have been adopted in this field, such as

the e-Government 2012 plan in Italy, which scheduled, beyond the simplification and digitalization

of digital prescriptions and disease certifications, online reservation systems, the creation of

infrastructures to supply healthcare services which meet consumers’ needs. In this national context

a technological infrastructure for a federated electronic patient record, namely Fascicolo Sanitario

Elettronico (FSE), that enables citizens and authorized health professionals to access the health

data wherever they are located in the national territory or abroad, preserving the citizen privacy,

and facilitating the management of the evolution of the process of care has been developed (within

the InFSE - Infrastruttura del Fascicolo Sanitario Elettronico - and the OpenInFSE - Evoluzione e

8 Ontologies, defined as “explicit specification of a conceptualization” [12], capture the meaning of a particular subject domain that corresponds to

what a human being knows about that domain. They are typically represented as classes, properties, attributes and values. Furthermore, ontologies

allow sharing knowledge between people, agents, and software; enable reuse of domain knowledge; and enable automated reasoning on data.


13

interoperabilità tecnologica del Fascicolo Sanitario Elettronico - projects) [12, 16]. Within the FSE

project some attempts have been made on the development of a specific section of FSE, namely

Taccuino personale del Cittadino, addressed to healthcare citizens for the easy registration of

personal and familiar healthcare information and data, on their habits concerning food and physical

activities, as well as for the registration of clinical documents issued by healthcare facilities that are

not affiliated to the National Healthcare System and to maintain a daily diary for relevant events

(e.g. visits, diagnostic exams, observable parameters monitoring, etc.). Different solutions have

been proposed by Regions and in the context of national research projects, but the most advanced

solution in terms of functionalities, interoperability and usability has been experimented and funded

by the Autonomous Province of Trento, where the TreC (Cartella Clinica del Cittadino) PHR has

become a concrete integrated service for the citizens of this territory and is the sole relevant case

in Italy where consumer-oriented vocabularies have been applied [44].

Regarding the possibility to use a CMV as support tool for the retrieval of health care information

and literature during web searching both performed by lay persons or by physicians, some studies

proposed methods providing reformulation of consumers’ queries for better medical information

search returns [30]. Results improves if the CMV is mapped to UMLS and in particular to MeSH, as

used for searching purposes and health-related literature indexing) [23].


14

5 Conclusions

In this annex the problem of the linguistic gap between “lay” and “specialized” terminology in the

medical domain has been treated. The reviewed studies have shown that patients often create

semantic ambiguities in using lay terms for expressing clinical concepts, in fact they use some

terms interchangeably as synonyms (as in the case of headache and migraine) making errors.

Above all they find it difficult to understand technical medical terms used by physicians and other

professionals in clinical documents and EHRs as well as in informative webpages, etc. This can

wrongly influence medical information comprehension and consequently decision making. In order

to avoid these consequences researchers have started to address this issue through the creation

of consumer-oriented medical vocabularies by also providing their integration with standard and

international medical terminologies and classification systems used in the healthcare domain, in

contrast to traditional approaches which proposed the use of specialized medical terminologies to

be integrated in consumer-oriented healthcare application such as personal health records or

mobile application for the monitoring particular health conditions.

The present document highlights the need for mapped consumer-oriented vocabularies in order to

guarantee an integration framework that can be reused in different consumer-oriented healthcare

applications, taking advantage of Semantic Web technologies.

The application of consumer-oriented medical vocabularies can contribute to the support of

healthcare consumers and laypersons as well as physicians in different scenarios:

1) translating and interpreting clinical notes or test results containing medical jargon (e.g., mapping

between a standardized medical vocabulary used in the physician’s EHR to a consumer-oriented

vocabulary could be useful in providing consumer-understandable names to help patients interpret

these documents);

2) searching for healthcare information (e.g., facilitating automated mapping of consumer-entered

queries to technical terms – if the term queried is mapped to a thesaurus such as MeSH it would

produce better search results during the search in a bibliographic database such as PubMed);

3) helping patients in the description of their clinical history, symptoms and complaints both in their

PHR and in online medical consultations; helping physicians and other healthcare providers in the

process of encoding reasons for encounters (symptoms, diseases, diagnoses and procedures);

helping physicians to automatically interpret their patients clinical history stored in their PHR, and

to automatically produce clinical notes understandable by healthcare consumers and patients.

Finally, many potential applications in the patient empowered health care (e.g. Home care) are

possible.


15

References

1. Abdaoui A., Azé J., Bringay S., Grabar N., and Poncelet P. Analysis of forum messages

written by health professionals and patients In Proceedings of MIE 2014. Istanbul,

Turkey. Poster.

2. Baldini M., Parlare al paziente, parlare “col” paziente, in Proceedings of the Conference

L’arte medica tra comunicazione, relazione, tecnica e organizzazione, Scriptorium,

Torino, pp. 9-25, 1996.

3. Baorto D. M. and Cimino J. J.. An “infobutton” for enabling patients to interpret on-line

pap smear reports. In Proceedings of AMIA Symposium - AMIA2000, pages 47–50,

2000.

4. Berners-Lee T., Hendler J., and Lassila O.. The Semantic Web. Scientific American,

284(5), 34—43, 2001.

5. Bodenreider O.. Issues in Mapping LOINC Laboratory Tests to SNOMED CT. In AMIA

Annual Symposium, AMIA2008, pp. 51—55, 2008.

6. Cardillo E.. A lexi-ontological resource for consumer healthcare. The Italian Consumer

Medical Vocabulary, Doctoral Thesis, Fondazione Bruno Kessler, University of Trento,

Italy, April 2011.

7. Cardillo E., Chiaravalloti M. T., Pasceri E.. Assessing ICD-9-CM and ICPC-2 Use in

Primary Care. An Italian Case Study. In Proceeding of the 5th International Conference

on Digital Health (Digital Health 2015), Florence, 18-20 May 2015 (accepted).

8. Cardillo E., Hernandez G., and Bodenreider O.. Integrating consumer-oriented

vocabularies with selected professional ones from the UMLS using Semantic Web

Technologies. Proceedings of the 3rd International Conference of Electronic Health

(eHealth2010), 12-15 December, Casablanca, 2010.

9. Cardillo E., Tamilin A., and Serafini L.. A Methodology for Knowledge Acquisition in

Consumer-oriented Healthcare. In Knowledge Discovery, Knowledge Engineering and

Knowledge Management: IC3K Revised Selected Papers, pp. 64 -71, Springer Berlin,

2010.

10. Cardillo E., Warnier M., Roumier J., Jamoulle M., Vander Stichele R.. Using ISO and

Semantic Web standards for creating a Multilingual Medical Interface Terminology: A

use case for Heart Failure, in Guadalupe Aguado de Cea, Nathalie Aussenac-Gilles,

Terminologies and Terminology and Artificial Intelligence, Paris, IRIT, 2013, pp. 27-34,

(10th International Conference on Terminology and Artificial Intelligence - TIA2013,

Paris, 28 - 30 October).

11. Ceusters W., Smith B. and De Moor G.. Ontology-Based Integration of Medical Coding

Systems and Electronic Patient Records. In MIE2005, 2005.

12. Chiaravalloti M. T., Ciampi M., Pasceri E., Sicuranza M., De Pietro G., Guarasci R.. A

model for realizing interoperable EHR systems in Italy, in Proceedings of the 15th

International HL7 Interoperability Conference, pp. 13-22.

13. De Clercq P. A., Hasman A., and Wolffenbuttel B. H.. A consumer health record for

supporting the patient-centered management of chronic diseases. Medical Informatics

and the Internet in Medicine, 28(2):117–127, June 2003.

14. Dell’Orletta F., Venturi G., and Montemagni S.. Unsupervised Linguistically-Driven

Reliable Dependency Parses Detection and Self-Training for Adaptation to the

Biomedical Domain. In Proceedings of the 2013 Workshop on Biomedical Natural

Language Processing (BioNLP 2013), Association for Computational Linguistics, pages

45–53, Sofia, Bulgaria, August 4-9, 2013.


16

15. Doing-Harris K. M. and Zeng-Treitler Q.. Computer-assisted update of a consumer

health vocabulary through mining of social network data. J Med Internet Res. 2011 May

17;13(2):e37.

16. Esposito A., Sicuranza M., Ciampi M.. A patient centric approach for modeling access

control in EHR systems. 13th International Conference on Algorithms and Architectures

for Parallel Processing, ICA3PP 2013. Springer International Publishing, pp. 225-232.

17. Euzenat J. and Shvaiko P.. Ontology Matching. Springer, 2007.

18. Flatley B. P. et al.. Project HealthDesign: Rethinking the power and potential of personal

health records. Journal of Biomedical Informatics , Volume 43 , October 2010, Issue 5 ,

S3 - S5.

19. Grabar N. and Hamon T.. Automatic extraction of layman names for technical medical

terms. In Proceedings of ICHI 2014. Pavia, Italy.

20. Gruber M., Thomas R.. Toward principles for the design of ontologies used for

knowledge sharing. Padua workshop on Formal Ontology, 1993.

21. Heymans Institute of Pharmacology EEC. Multilingual Glossary of Technical and

Popular Medical Terms in nine European Languages: Final report, Technical report,

University of Ghent, Mercator College, Department of Applied Linguistics, Ghent,

Belgium, 1995.

22. Hong Y., Ehlers K., Gillis R., Patrick T., and Zhang J.. A Usability Study of Patient-

friendly Terminology in an EMR System. In Studies in Health Technology and

Informatics, Volume 160: MEDINFO 2010; pp. 136 – 140. doi=10.3233/978-1-60750-

588-4-136.

23. Hong Y., Gillis R. D., Donnell R. F. .Use of consumer health vocabularies in online

physician directory to improve physician search. In Proceedings of AMIA Annu Symp

2008, Nov 6:974.

24. Kandula S., Curtis D., and Zeng-Treitler Q.. A Semantic and Syntactic Text

Simplification Tool for Health Content. In Proceedings of AMIA Annual Symposium, pp.

366-370, 2010.

25. Keselman A., Logan R., Smith C.A., Leroy G., and Zeng Q., Developing Informatics

Tools and Strategies for Consumer-centered Health Communication, in «Journal of the

American Medical Informatics Association», 2008, vol. 14, n. 4, pp. 473-483.

26. Keselman A. and Smith C.A.. A classification of errors in lay comprehension of medical

documents. J Biomed Inform. 2012 Dec;45(6):1151-63.

27. Kim H., Zeng Q., Goryachev S., Keselman A., Slaughter L., and Smith C.A.. Text

Characteristics of Clinical Reports and Their Implications for the Readability of Personal

Health Records, in Kuhn K. A., Warren J. R., Leong T.-Y. (eds.): MEDINFO 2007 -

Proceedings of the 12th World Congress on Health (Medical) Informatics. Building

Sustainable Health Systems, 20-24 August 2007, Brisbane, Australia, in «Studies in

Health Technology and Informatics», 2007, vol. 129, pp. 1117-1121.

28. Jamoulle M., Cardillo E., Roumier J., Warnier M, Vander Stichele R.. Mapping French

terms in a Belgian guideline on heart failure to international classifications and

nomenclatures: the devil is in the detail. In Informatics in Primary Care (2014), 4, pp.

189-198. DOI=10.14236/jhi.v21i4.66

29. Jeongeun K., Jeeyoung J. and Yoonju S.. An Exploratory Study on the Health

Information Terms for the Development of the Consumer Health Vocabulary System. In

Studies in Health Technology and Informatics; Volume 146: Connecting Health and

Humans; Ios Press, 785 – 785, 2009. doi=10.3233/978-1-60750-024-7-785.


17

30. Jiang L. and Yang C. C.. Using Co-occurrence Analysis to Expand Consumer Health

Vocabularies from Social Media Data. IEEE International Conference on Healthcare

Informatics (ICHI), 2013.

31. Iandolo C., Parlare col Malato. Tecnica, Arte ed errori della comunicazione. Armando,

Roma, 1983.

32. Lefever, E., Macken, L., & Hoste, V. (2009). Language-independent bilingual

terminology extraction from a multilingual parallel corpus. Proceedings of the 12th

Conference of the European Chapter of the Association for Computational Linguistics.

The Association for Computational Linguistics, Athens, Greece.

33. Leroy G., Endicott J. E., Mouradi O., Kauchak D., and Just M. L.. Improving perceived

and actual text difficulty for health information consumers using semi-automated

method. AMIA Annu Symp Proc. 2012; pp. 522-531, 2012.

34. Lucchini A. (eds.). Il linguaggio della Salute. Come migliorare la comunicazione con il

paziente, Sperling & Kupfer, Milano, 2008.

35. Marshall P. D., Bridging the Terminology Gap Between Health Care Professionals and

Patients with the Consumer Health Terminology (CHT), in Proceedings of AMIA Annual

Symposium, AMIA2000, p. 1082, 2000.

36. Messai R., Simonet M., Bricon-Souf N., and Mousseau M.. Characterizing Consumer

Health Terminology in the Breast Cancer Field. In Studies in Health Technology and

Informatics, Volume 160: MEDINFO2010, pp. 991 – 994, 2010. doi=10.3233/978-1-

60750-588-4-991.

37. Ofoghi B., Lopez-Campos G. H., Martin Sanchez F. J., and Verspoor K.. Mapping

Biomedical Vocabularies: A Semi-Automated Term Matching Approach. In Studies in

Health Technology and Informatics, Volume 202: Integrating Information Technology

and Management for Quality of Care, pp. 16 – 19. 2014. doi=10.3233/978-1-61499-423-

7-16

38. Polepalli R. B., Houston T., Brandt C., Fang H., and Y. H.. Improving patients' electronic

health record comprehension with NoteAid. Stud Health Technol Inform. 2013;192:714-

8.

39. Plovnick R. M. and Zeng Q. T.. Reformulation of Consumer Health Queries with

Professional Terminology: A Pilot Study. In J Med Internet Res 2004;6(3):e27.

DOI=10.2196/jmir.6.3.e27

40. Seedor M., Peterson K. J., Nelsen L. A., Cocos C., McCormick J. B., Chute C. G., and

Pathak J.. Incorporating expert terminology and disease risk factors into consumer

health vocabularies. Pac Symp Biocomput. 2013:421-32.

41. Smith C. A.. Consumer language, patient language, and thesauri: a review of the

literature. J Med Libr Assoc. 2011 Apr;99(2):135-44. DOI= 10.3163/1536-

5050.99.2.005.

42. Smith C. A., Hetzel S., Dalrymple P., and Keselman A.. Beyond readability:

investigating coherence of clinical text for consumers. J Med Internet Res. 2011 Dec

2;13(4):e104.

43. Soergel D., Tse T., and Slaughter L., Helping Healthcare Consumers Understand: an

“Interpretative Layer” for Finding and Making Sense of Medical Information, in

Proceedings of the international conference IMIA2004, San Francisco, California, pp.

931-935, 2004.

44. TreC – Cartella Clinica del Cittadino. The project is consultable at:

http://trentinosalute.net/trec

45. Tudorache T., Nyulas C. I., Noy N. F., Redmond T., and Musen M. A.. iCAT: A

Collaborative Authoring Tool for ICD-11. In Proceedings of the ISWC 2011 Workshop

http://trentinosalute.net/trec


18

Ontologies Come of Age in the Semantic Web (OCAS), CEUR-WS Vol-809, Bonn,

Germany, October 24th, 2011.

46. Zeng Q., Goryachev S., Keselman A., and Rosendale D.. Making Text in Electronic

Health Records Comprehensible to Consumers: A Prototype Translator, in Proceedings

of the AMIA Annual Symposium – AMIA2007, 2007, pp. 846-850.

47. Zeng Q. and Tse T.: Exploring and Developing Consumer Health Vocabulary. In JAMIA

2006;13:24-29

48. Zeng Q., Tse T., Divita G., Keselman A., Crowell J., Browne A. C., Goryachev S., and

Ngo L., Term Identification Methods for Consumer Health Vocabulary Development, in

«Journal of Medical Internet Research», 2007, vol. 9, n. 1, p. 4.

49. Zeng-Treitler Q., Kandula S., Kim H., and Hill B.. A Method to Estimate Readibility of

Health Content. ACM SIGKDD Workshop on Health Informatics (HI-KDD 2012),

Beijing, China, 2012.

50. Zhang S., and Bodenreider O.. Experience in aligning anatomical ontologies.

International Journal on Semantic Web and Information Systems, 2007, 3(2):1-26.

51. WellMed Medical Management, <https://www.wellmedmedicalgroup.com/>.

52. Widdows D., Peters S., Cederberg S., Chan C-K.. Unsupervised Monolingual and

Bilingual Word-Sense Disambiguation of Medical Documents using UMLS. In

Proceedings of the ACL 2003 workshop on Natural language processing in biomedicine

- Volume 13 (BioMed '03), Vol. 13. Association for Computational Linguistics,

Stroudsburg, PA, USA, 9-16. doi=10.3115/1118958.1118960

53. Wu D. T., Hanauer D. A., Mei Q., Clark P. M., An L. C., Lei J., Proulx J., Zeng-Treitler

Q., and Zheng K. Applying multiple methods to assess the readability of a large corpus

of medical documents. Stud Health Technol Inform. MEDINFO2013, vol. 192, pp.647-

51, 2013. doi:10.3233/978-1-61499-289-9-64

Medical terminologies for patients - SemanticHealthNet 288408... · Medical terminologies for...

Documents

Transcript of Medical terminologies for patients - SemanticHealthNet 288408... · Medical terminologies for...